The impact of an independent response evaluation committee (REC) using RECIST guidelines in a Four-Arm Cooperative Study (FACS) for advanced non-small cell lung cancer (NSCLC) in Japan

Abstract

7065 Background: From 44 hospitals in Japan, 602 patients (pts) with advanced NSCLC were registered in FACS. In 320 cases achieving more than 20% decrease in tumor size according to RECIST guidelines, a blinded peer review was prospectively performed by an independent REC. The purpose of the study was to investigate the impact of the independent REC using RECIST guidelines. Methods: Two diagnostic radiologists and two medical oncologists identified target and nontarget lesions from baseline medical images. Target lesions were measured by calipers. The concordance and discordance between the investigators and the REC were assessed. Using the Fisher's exact test, attempts were made to identify the factors for discordance, such as best overall response by investigators, alterations of target and nontarget lesions by REC and the number of pt accruals by institution. Results: Four pts were excluded due to insufficient data. Target and nontarget lesions documented by investigators were altered in 109 (34.5%) of 316 pts. The best overall response between investigators and REC are given in the table below. Of 209 PR pts, 52 pts were reclassified to SD and 3 pts reclassified to PD. Of 106 SD pts, 30 pts were reclassified to PR and 7 pts reclassified to PD. The Kappa value was 0.40 (95% CI (confidence intervals); 0.30 - 0.50) and the proportion of agreement was 70.9% (95% CI; 65.9 - 75.8). The number of PR cases reduced by 12.0% after review of the REC. There were no significant factors for discordance between investigators and REC. Conclusions: Despite both groups using RECIST guidelines, inter-observer variability between investigators and the independent REC existed. Investigators had a tendency to overestimate tumor response to chemotherapy, however, in SD cases judged by investigators, had an underestimation. No significant financial relationships to disclose.